Data Mining Applications in a Medical System: A Case Study

Data Mining Applications in a Medical System: A Case Study

Morteza Bagherpour (Department of Industrial Engineering, Iran University of Science and Technology, Tehran, Iran), Asma Erjaee (Pediatrics Department, Shiraz University of Medical Sciences, Shiraz, Iran), Amir Hossein Rasekh (Department of Computer and Electrical Engineering, Shiraz University, Shiraz, Iran) and Seyed Mohsen Dehghani (Pediatrics Department, Shiraz University of Medical Sciences, Shiraz, Iran)
Copyright: © 2014 |Pages: 7
DOI: 10.4018/978-1-4666-5202-6.ch055

Chapter Preview


Data Collection

The patient related data was gathered through a randomized clinical trial study, where all children <18 year of age with possibility of H. pylori infection; according to their sign and symptoms, whom had referred to the Gastrointestinal clinic afflicted to Shiraz University of Medical Sciences from April 2011 till September 2011 were enrolled. First a questionnaire form was completed for each patient, including questions regarding the patient’s symptoms (e.g. abdominal pain, nausea, vomiting, halitosis, GI bleeding …) there duration, positive history of treatment with antacids (H2 blockers and proton pump inhibitors), and any positive family history of acid peptic diseases in their first degree relatives. Also all patients were examined for tenderness in there epigastric area and if so this was entered in the form. The patient’s weight and height were as well recorded in the questionnaire form. Questions regarding symptoms which could be possibly correlated to H.pylori infection in children were derived from previous studies on this concept (Drumm, 1993 and Giacomo et al., 2002 and Gold et al., 2000).

Further an endoscopy was performed for all subjects, through which an antral and corpous mucosal biopsy was obtained for histopathology and RUT. Biopsy specimens for histology were fixed in formalin and were sent to Shahid Motahari Pathology Laboratory of Shiraz University of Medical Sciences for analysis. Results regarding the histopathology and RUT were also entered in the form.

An informed consent was obtained from parents of all patients. The used dataset is collected through a six months period from those patients who need to perform the Endoscopy in order to diagnose the existence of H.pylori infection. The features of dataset are:

Key Terms in this Chapter

Accuracy: One minus error percentage.

Low Cost of Diagnosis: Cheap methods to detect infection with low costs.

Bayesian Learning: Bayesian learning methods is able to provide useful practical solutions and forecasting features toward solving complicated problems.

Decision Tree: Decision trees are useful for topics that can be given in the form of unit answers such as category or class name

Endoscopy: Endoscopy is an invasive procedure, and is not always accessible; moreover the high costs of this test may be another reason that may lead us to consider other possible alternatives for diagnosis of this infection.

H. Pylori: Helicobacter pylori is a type of infection may occur in children.

Data Mining: Analysis step of the “Knowledge Discovery in Databases” process.

Logistic Regression: Logistic regression is a particular type of regression which is used in cases that response variable is double-choice or multiple-choice.

Complete Chapter List

Search this Book: